David H. Ledbetter, Ph.D.

david.ledbetter@emory.edu

ISCA: International Standard CytogenomicArray (ISCA) Consortium and Database

July 8, 2009

1981“Cytogenetics will become extinct within the next 5 years.”

C. Thomas Caskey, M.D., Chair Department of Human GeneticsBaylor College of Medicine

Cytogenetics: The FIRST whole genome technology!

Requires 500-600 evenly spaced DNA probes to match the power of the karyotype.

From N. Chia

Key Features of G-banded Karyotype

• First whole-genome technologyto detect clinically significantgenomic imbalances (deletions, duplications)

• Benign polymorphisms (CNVs)identified by empiric experience over a number of years

From N. Chia

Gene dosage lessons from 50 years of cytogenetics experience

• Monosomy and deletions cause more severe phenotypicconsequences than trisomy and duplications

• No viable autosomal monosomies (only 45,X)

• Larger imbalances (more genes) more severe phenotype thansmaller imbalances

• Imbalance of G-negative bands (gene-rich) more severe thanG-positive bands (gene-poor)

From N. Chia

Gene dosage lessons from 50 years of cytogenetics experience

• Not all genes are dosage-sensitive• Down syndrome “critical region”

• phenotype in microdeletion syndromes attributed to 1 or few genes (UBE3A -> Angelman syndrome)

From N. Chia

Key Features of G-banded Karyotype

• Clinical significance of imbalance in proband sometimesrequires parental studies to determine if pathogenic or benign(de novo taken as evidence likely pathogenic)

But, limited resolution (5-10 Mb), variable quality and subjective interpretation

Lesson 1: The “Gold Standard” karyotype has become tarnished

From N. Chia

How much structural variation is there in humans?-individual and population (*note vast majority is CNV)

http://projects.tcag.ca/variation/

Iafrate et al, Nature Genetics 2004

Gene Dosage Map and CNVs

How many genes in the genome are dosage sensitive?(haploinsufficiency or triplosensitive)• Probably a minority (? 5-10%).• Many genes are not dosage sensitive

• heterozygous carriers for autosomal recessive disorders

? 10 CNVs or 1 dosage insensitive region with an infinite # of possible CNVs

Evolution of Array Designs0 +1-1

Targeted

WMS

tel

tel

cencen

Targeted + 850 BAC or oligo array

0 +1-1

>3 Mb gaps

Targeted + 1 Mb or 500 kb

0 +1-1 0 +1-1

Targeted + Whole Genome

Chr 7

0.5 -1 Mb gaps

Targeted + Whole Genome Arrays

Oligonucleotide microarray (60mers)

Custom-designed 4x44k format - Agilent

Telomere FISH cloneUnique centromere FISH clone

cen qterpter

Known clinically relevant regions

del/dup del/dup

75 kb interval backbone

Custom Array Design by Clinical Cytogeneticists & Clinical Geneticists

~250 (500) kb

~50 kb

Resolution

Baldwin et al., 2008

Why Use a 500kb Triage?

CNVs Based on Size

Size Group (kb)

0102030405060708090

100

1-499 500+

~95%

~5%

% C

NVs

per

Gro

up

Redon et al. (Nature 2006) – 81 kb median with 500K arrayLee et al. (unpublshed) – 2.7 kb median with 4.2 M array

Database of Genomic Variants (Oct. 2006)(http://projects.tcag.ca/variation/

Pathogenic vs. Benign Copy Number Changes

1. Region of known clinical significance:• known del/dup or Mendelian disorders• known benign CNC • comparison with other cases in literature,

databases

2. Gene Content • correlates with size and location

(G- bands gene-rich; G+ gene-poor)

3. Inherited or de novo (need parental samples in

Targeted Coverage: PWS/AS Region

PWS/AS deletion

Targeted Coverage: PWS/AS Region

PWS/AS deletion Atypical deletion

UBE3A45 kb loss

Whole Genome Coverage

12q: 4.7 Mb deletion ~11 known genes

Case 1

15q: 4.5 Mb deletion~21 known genes

Case 2

2p: 3.0 Mb deletion~ 12 known genes

Case 3

del 2

RP11-426D14

FISH Confirmation – 2p deletion

nl 2

32K BAC set

FISH = mech.

Targeted vs. Whole Genome Detection Rates

Abnormal detection rate: 18%

Targeted coverage: 13%

Whole genome coverage: 5%

Whole genome coverage enhances the detection of clinically relevant cytogenetic imbalances

To date, more than 3,000 cases analyzed…

10% of patients who have karyotype first have a significantly delayed diagnosis!

Referring Dx:Dysmorphic featuresDevelopmental delayHypotoniaHypoplastic penis

17p: 2.3 Mb deletion

Case 11

Case 11 Loss of 17p13.2p13.1: ~2.3Mb

Referring Dx:Dysmorphic featuresDevelopmental delayHypotoniaHypoplastic penis

17p: 2.3 Mb deletion

p53 loss = Li-Fraumeni syndrome, high cancer risk

Adam et al., J Pediatrics, Jan., 2009Other cases: RB1, VHL, Peutz-Jeghers

Cancer Susceptibility

Mechanisms of Chromosome Rearrangements

Terminal telomere deletions with adjacent duplications –pre-meiotic breakage-fusion-bridge cycles after random breakage

16 cases:2q x 24p5p6q8p9p x 310q15q18p18q22q x 3

Mechanisms of Chromosome Rearrangements

Breakpoint analysis: Random or specific mechanism?

Examined 54 cases with copy number imbalances (300 kb-10 Mb in size) with known inheritance:

15% mediated by flanking segmental duplications (NAHR)

85% were not associated with seg dups and most likely represent random chromosome breakage

C. Lee et al.: 7% of CNVs are associated with NAHR; majority are random

Current Status of Cytogenetic Array Testing

• Multiple platforms• BAC vs. oligo • aCGH, SNP, beadchipAll detect single copy loss and gain accurately

• Variable design and content• Targeted + whole-genome• increasing number of clinical loci including

Mendelian genes

• ~300 cyto labs in U.S.• ? need/want 300 aCGH designs

June 23-24, 2008 (Atlanta, GA)30 attendees from U.S., Canada,

UK and BrazilClinical Geneticists, Clinical Molecular &Cytogeneticists, Genomics & Bioinformatics

December 15-16, 2008 (Bethesda, MD)60 attendees from U.S., Canada, UK, Belgium, Netherlands, Italy, Brazil 5 industry reps (Affymetrix, Agilent, BlueGnome, Nimblegen, OGT)NCBI, NHGRI, NIMH, NICHD

International aCGH Workshops(https://isca.genetics.emory.edu)

Summary of 1st workshop(https://isca.genetics.emory.edu)

• Central, public database for clinical cyto array data (raw data files and normalized data) extremelyvaluable to clinical and research communities to rapidly identify pathogenic vs. benign CNCs

• all de-identified data to achieve largest numbers, albeit with minimal clinical info

• complete raw data and normalized data files

• encourage informed consent and detailed clinical information for DECIPHER submission whenever possible

Summary of 1st workshop(https://isca.genetics.emory.edu)

• Need more, high quality data on benign CNCs in normal controls, including mutation rate

• Consensus that cytogenetic array should be 1stline diagnostic test for unexplained MR, MCAinstead of karyotype (D. Miller, ms. in prep.)

• Need expert committee and evidence-based standards to make recommendations re:• clinical indications for testing• minimum standards for design, content,

resolution, QA/QC• guidelines for interpretation and reporting

2nd workshop(https://isca.genetics.emory.edu)

• New, higher quality data on normal controls from research community; culling of poor data from DGV

• NCBI received NIH IRB approval for de-identified data submission to dbGaP using “opt-out” mechanismof consent

• Increased international participation (Canada, UK, Netherlands, Belgium, Italy)

Leslie Biesecker (NHGRI/NIH) Nigel Carter (Sanger Institute, UK) John Crolla (Salisbury, UK) Evan Eichler (University of Washington) Ada Hamosh (Johns Hopkins/OMIM) David Ledbetter (Emory University) Charles Lee (Harvard-Brigham & Women’s) Christa Martin (Emory University) David Miller (Harvard-Boston Children’s) Nancy Spinner (CHOP)Joris Vermeesch (Universiteit Leuven, Belgium)Greg Peters (Australia)

ISCA Steering Committee(https://isca.genetics.emory.edu)

International Public Database for Cytogenomic Array Data

• Initially, minimal phenotypic data require-ment but efforts to encourage detailed phenotypic data and submission to DECIPHER

• Will perform quality checks, summary tables,and public data release on quarterly basis

• available to UCSC, Ensembl, DECIPHER,commercial vendors, local lab databases

Proposal for a public database and evidence-based guidelines for design and interpretation

• Technology platform and vendor neutral: BAC, oligo, beadchip• Common denominator is genome sequence

coordinates for gains and losses

• Develop evidence-based guidelines for optimal design and interpretation• Minimum standards

Alberta Children’s HospitalARUP/University of UtahBeth Israel Deaconess Medical CenterChildren’s Hospital of PhiladelphiaChildren’s Memorial Hospital, ChicagoCincinnati Children’s HospitalCredit Valley HospitalDuke UniversityEmory UniversityGeneDxHamad Medical Corporation, QatarHenry Ford HospitalHospital for Sick Children, TorontoKaiser Regional Cytogenetics LabLondon Health Sciences CentreMayo ClinicMission Health, Fullerton Genetics Lab

Current members of the Consortium:(agreed to public data sharing)

Montefiore HospitalNorthwestern Reproductie Genetics, ChicagoStanford Hospitals and ClinicsSudbury Regional HospitalTexas Tech UniversityU. Mass. Memorial Medical CenterUMCG, Groningen, NetherlandsU. Alabama, BirminghamU. FloridaU. MichiganU. NebraskaU. Oxford, UKU. RochesterU. Sao Paolo, BrazilU. WisconsinU. Medical Center, LjubljanaWessex Regional Genetics Lab

Current members of the Consortium:(agreed to public data sharing)

ISCA “Community” array design (for labs that don’t have own custom designs)Current array – 44k (4-plex), 105k (2-plex)

ISCA drafts – 180k (4-plex)

140k assigned; 40k available for customization

Result of merging designs of existing arrays: Emory – Ledbetter/Martin GeneDx – Aradhya Salisbury, UK – Crolla/Barber Oxford, UK – Knight/Smith/Connell Dutch Consortium/Oxford design - Kok Belgium Consortium - Vermeesch

…and continued improvements based on recommendations from ISCA Steering Committee

Telomere FISH cloneUnique centromere FISH clone

cen qterpter

Clinically relevant targets (~500)

Target Target

~25, 35 or 75 kb interval backbone(corresponds to 180K, 105K, 44K)

~100 - 250 kb

~30 - 50 kb

Resolution

Whole-genome plus Targeted Community Array Design

Backbone: 2q24chr2:157,747,500-161,647,500 basepairs

44K105K180K

44K: ~75 kb spacing (225 kb resolution)

105K: ~35 kb spacing (105 kb resolution)

180K: ~25 kb spacing (75 kb resolution)

Targeted Gene: UBE3Achr15:23,105,326-23,250,028 size: 145 kb

44K105K180K

Targeted Gene: MECP2chrX:152,921,476-153,035,363 size: 114 kb

44K105K180K

Summary of Consortium Experience• Whole-genome oligo array clinical testing

implemented April, 2007 (Emory and GeneDx)

• Over 25,000 clinical cases performed to date; Currently >500 cases/week.

• del 16p11.2 most common finding (1/300)• 1-2 new cases del 16p11.2 identified each week

del 16p11.2 and autism

del 16p11.2: clinical aCGH cases

Age range

Num

ber o

f pat

ient

s

Model for Genotype -> Phenotype Studies

Patients

Identified on a clinical basis (“free”)

Early identification of at-risk children

Emory: 10 patients

GeneDx: 6 patients

1/300 clinical aCGH tests = del 16p11.2

at least one new patient/week in consortium

Conclusions & Predictions

1) Postnatal Cytogenetics- aCGH is much more sensitive that G-banded karyotype for postnatal, pediatric applications and may soon become the primary genetic test for children with unexplained developmental delay, mental retardation, birth defects, seizures, autism, etc.

2) Prenatal Diagnosis-NIH sponsored multicenter trial on prenatal

aCGH underway to compare aCGH to G-banded karyotype. Results in ~2 years, but expect aCGH to win.

Acknowledgements

Emory University:Christa Lese Martin, Ph.D.Erin B. Kaminsky, Ph.D.Margaret Adam, M.D.

Grant Support:NIHLuminex-ACMGF

The physicians, genetic counselors and familiesinvolved in these cases.

How much structural variation is there in humans?�-individual and population (*note vast majority is CNV) Why Use a 500kb Triage?Targeted Coverage: PWS/AS RegionTargeted Coverage: PWS/AS RegionWhole Genome CoverageCase 11 Case 11 Cancer Susceptibility Whole-genome plus Targeted �Community Array DesignBackbone: 2q24Targeted Gene: UBE3ATargeted Gene: MECP2del 16p11.2 and autismdel 16p11.2: clinical aCGH casesModel for Genotype -> Phenotype Studies